Joint Spectral Clustering based on Optimal Graph and Feature Selection

Abstract

Redundant features and outliers (noise) included in the data points for a machine learning clustering model heavily influences the discovery of more distinguished features for clustering. To solve this issue, we propose a spectral new clustering method to consider the feature selection with the L 2,1 -norm regularization as well as simultaneously learns orthogonal representations for each sample to preserve the local structures of data points. Our model also solves the issue of out-of-sample, where the training process does not output an explicit model to predict unseen data points, along with providing an efficient optimization method for the proposed objective function. Experimental results showed that our method on twelve data sets achieves the best performance compared with other similar models.

IGRF-RFE: A Hybrid Feature Selection Method for MLP-based Network Intrusion Detection on UNSW-NB15 Dataset

“…In the prediction stage, after producing a class probability vector, the argmax function, as defined in equation 10, finds the largest number among them and returns its index. result = argmax(probability vector) (10) Our model was trained with Adam's optimization algorithm, which adaptively adjusts the learning rate based on recent gradients for the weight. Also, our model used the learning rate = 0.0003, the batch size = 64, and the epochs = 300.…”

“…classifies network behavior by learning from the labeled data [9]. Unsupervised intrusion detection methods such as K means and hidden Markov models focus on the clustering problem [10] to group network behaviors [11].…”

Artificial Intelligence-Enabled DDoS Detection for Blockchain-Based Smart Transport Systems

“…In contrast, we use supervised learning for the MLP model with labels to evaluate the classification accuracy. We first use autoencoder (AE) as a feature extraction tool [ 28 ] to find the most relevant features from the original dataset and then use multi-layer perceptron (MLP) as a classifier that categorizes the attacks into different categories (i.e., classes). The overview of the hybrid approach we use is depicted in Figure 3 .…”

Task-Aware Meta Learning-based Siamese Neural Network for Classifying Obfuscated Malware

“…Though it has proven that the embedded N-grams of opcodes Li et al [2020] and graph embedding Zhang et al [2020], Hashemi et al [2017], Zhu et al [2021] can efficiently capture unique malicious components, however, conducting the opcodes and graph embedding is time-consuming and their generalizability limited (e.g., the graph embedding tends only work well in learning the static features). Using these techniques, a unique behavior presented in a malware family that is critical to capture the variant of that malware family often tends to be regarded as noises while only common known behaviors of malware families are captured by the CNN model, for example, the work by Microsoft collaborated with Intel demonstrates the setting of the practical value of the image-based transfer learning approach for static malware classification Chen et al [2020].…”